EXPERIMENTAL INVESTIGATION OF CONCENTRATION DEPENDENT NON-IDEAL DIFFUSION IN HYDROCARBON
SYSTEMS by
© Ebubechi Azubuike Evulukwu A Thesis submitted to the School of Graduate Studies
In partial fulfillment of the requirements for the degree of
Masters of Engineering Faculty of Engineering
Memorial University of Newfoundland
August 2015
St. John’s Newfoundland and Labrador
i
ABSTRACT
Solvent extraction technology (Vapor extraction/VAPEX) has drawn a lot of industry attention due to its potential to be an alternative to Steam assisted Gravity Drainage (SAGD) in heavy oil production. However the mass transfer mechanisms involved is yet to be fully comprehended.
Reliable oil production rate data is scarce, hence the reluctance from oil companies to implement the technology on a field/commercial scale. More work is required at the experimental level to fully understand the intricacies of the technology and hence facilitate its commercialization.
Experiments were conducted to evaluate the one-dimensional diffusivity of butane solvent in Athabasca bitumen at varying temperatures. Given diffusion is driven by concentration gradient, the diffusivity cannot be assumed constant throughout the whole diffusion process. Hence the diffusivity was found as a function of butane solvent concentration (mass fraction). Diffusivity functions for ideal mixing and non-ideal mixing were computed. Butane vapor temperature (24.00oC and 34.65 psi) is kept constant while the bitumen temperature is varied at 5 levels (27.00oC, 30.25oC, 33.50oC, 36.75oC and 40.00oC).
Assuming ideal mixing between hydrocarbons in VAPEX experiments is prevalent in the field.
This is because finding a parameter in the solvent-bitumen mixing system that accounts for non- ideal mixing without upsetting the system is difficult. This work accounts for non-ideal mixing by constantly measuring the bitumen liquid hydrostatic pressure via pressure differential transmitters as diffusion occurs. With bitumen height change and amount of diffused butane solvent being monitored, the real density reduction (non-ideal density reduction) can be computed. Results showed that assuming ideal mixing over-estimates the density reduction. The deviation between ideal and non-ideal mixing density values increase as temperature increases.
This is supported by most literature in the field. As temperature tends to standard temperature (25.00oC), the effects of non-ideal mixing become insignificant.
A MATLAB model is used to predict the ‘bitumen growth’ (bitumen swelling), this is compared to ‘bitumen growth’ observed experimentally. The difference between the two (experimental –
ii
predicted) is minimized by optimizing the diffusivity function coefficients. Results showed that the diffusion values (obtained via diffusivity functions) decreases as temperature increases.
There was no ‘live oil’ drainage in this experiment so diffusion is governed by the butane solvent solubility in the bitumen. This butane solvent solubility decreases with increasing temperature.
At equal mass fractions (ωs) all non-ideal mixing diffusivity functions yielded higher diffusion values than ideal mixing diffusivity functions. This is logical because diffusion is driven by concentration gradient. Ideal mixing scenarios over-estimate density reduction on mixing and hence provide a smaller concentration gradient compared to non-ideal mixing. The assumption of ideal mixing conditions clearly underestimates the real diffusivity values. The deviation between ideal and non-ideal diffusivity functions also increased as temperature increased. This follows the same trend as the deviation between ideal and non-ideal mixing density results.
A macroscopic mass balance was used to independently validate the diffusivity functions. This mass balance predicted the change in solvent height after ‘bitumen growth’ had been resolved for the full experimental time. This is an independent validation because change in solvent height data was not used to obtain the diffusivity functions. All but one of the diffusivity functions (40.00oC) was independently validated. Lack of validation in the 40.00oC run was due to technical issues while running the experiment. For all validation data, the non-ideal diffusivity functions provided a better fit for the experimental data than the ideal diffusivity functions.
Finally, the experimentally determined butane slope decrease, bitumen slope increase and non- ideal mixing coefficients for all varied temperature conditions were used as input values to make models in Design Expert (DE). These models were used to predict the aforementioned parameters at a random bitumen temperature (28.50oC). An extra experiment was run at this temperature (28.50oC) to compare the diffusivity functions coming from the experimental data to those from (DE) predictions. The DE originated diffusivities functions showed a good fit with the experimentally originated diffusivity functions. The model is therefore a robust model and can be used to predict diffusivity of butane at 24.00oC, within a given bitumen temperature range of 27.00oC - 40.00oC, while also accounting for non-ideal mixing and concentration dependency.
iii
ACKNOWLEDGEMENTS
I would first like to thank the Lord God almighty for his guidance. Completing this thesis easily ranks as the most challenging thing I have done to date in my life. So many times I thought I would never be able to complete this work, yet he also somehow gave me the strength and motivation to keep plugging away.
I would also like to thank my family (Mommy, Daddy, Chike, Ugo and Iheanyi), who consistently served as my support system while completing this thesis. I felt a jolt of motivation after every phone call, email and skype video conversation with them.
I would like to thank Lesley (my professor) for her guidance along with her team; Edison, Shervin, Mohammed, Xiaolong and Kim. Would also like to thank the guys from Technical services and Brian at the glass blowing shop for all the help they gave me with putting my equipment together.
I hope this work serves as an inspiration to many and contributes to the eventual development of solvent extraction technology on a commercial scale.
iv
TABLE OF CONTENTS
LIST OF TABLES --- vii
LIST OF FIGURES --- viii
APPENDIX --- xi
NOMENCLATURE --- xii
1) INTRODUCTION--- 1
1.1. World Heavy Oil Overview --- 2
1.2. Canadian Perspective --- 5
1.3. In-situ Extraction Methods for Bitumen --- 8
1.3.1. Cyclic Steam Stimulation --- 9
1.3.2. SAGD --- 11
1.3.3. VAPEX --- 14
1.3.4. N-Solv --- 16
1.3.5. Hybrid Processes --- 17
1.4. Technological Impacts --- 19
1.5. Scope of Research --- 21
2) LITERATURE REVIEW --- 23
2.1. History of VAPEX Experiments --- 23
2.2. Viscosity correlations --- 46
2.3. Diffusivity/Diffusion coefficient --- 52
2.2.1. Constant diffusion coefficient methods --- 54
2.2.1.1. Pressure decay methods --- 54
2.2.1.2. Dynamic Pendant Drop Shape Analysis (DPDSA) --- 59
2.2.1.3. Computer Assisted Tomography (CAT) --- 60
2.2.1.4. Nuclear Magnetic Resonance (NMR) --- 65
2.2.2. Mutual diffusion coefficient --- 68
2.2.3. Concentration dependent diffusivity methods --- 70
2.4. Non-ideal mixing of solvent and oil --- 78
2.5. This work --- 85
3) EXPERIMENTAL METHODOLOGY --- 88
v
3.1. Experimental overview --- 88
3.2. Equipment set-up --- 89
3.2.1. Butane Loading --- 90
3.2.2. Bitumen Loading --- 92
3.2.3. Solvent gas phase purity and vacuum creation --- 96
3.2.4. Run procedure --- 101
3.2.5. Troubleshooting--- 103
4) NUMERICAL METHODOLOGIES --- 104
4.1. Bitumen/Butane Density and Pressure Calculations --- 104
4.1.1. Non-ideal mixture density --- 104
4.1.2. Ideal mixture density --- 107
4.2. Bitumen height analysis --- 108
4.3. Solvent Continuity and Diffusivity Calculations --- 115
4.3.1. Diffusivity equations --- 115
4.3.2. Accounting for non-ideal mixing --- 116
4.3.3. Butane solvent solubility in bitumen --- 121
4.4. Using Design Expert --- 123
5) DISCUSSION OF RESULTS --- 126
5.1. Butane solvent height changes as a function of time --- 126
5.2. Bitumen height changes as a function of time --- 130
5.3. Ideal & non-ideal mixing density results --- 135
5.4. Diffusivity results --- 140
5.4.1. Diffusivity functions --- 141
5.4.2. Butane solvent mass fraction profile and bitumen density profile --- 142
5.4.3. Ideal and non-ideal mixing comparisons --- 144
5.4.4. Temperature comparisons --- 153
5.4.5. Diffusivity results validation--- 159
5.5. Design expert correlation results --- 165
5.5.1. Butane solvent decrease results --- 165
5.5.2. Bitumen increase results --- 169
vi
5.5.3. Non-ideal mixing results --- 172
5.5.4. Diffusivity results and validation --- 174
5.6. Experimental Error --- 181
6) CONCLUSIONS AND RECOMMENDATIONS --- 188
6.1. Conclusions --- 188
6.2. Recommendations --- 189
7) REFERENCES --- 191
APPENDIX --- 198
APPENDIX A - Bitumen assay characteristics --- 198
APPENDIX B - Troubleshooting --- 201
APPENDIX C - Starting experimental values --- 207
APPENDIX D - Bitumen/Butane density and pressure calculations --- 207
APPENDIX E - Bitumen height analysis table --- 215
APPENDIX F - Specific volume graphs (all temperatures) --- 216
APPENDIX G - Butane/Bitumen changes as function of time calculations --- 218
APPENDIX H - Design expert ANOVA tables --- 222
APPENDIX I - Matlab code & diffusivity function results --- 223
APPENDIX J - Solvent concentration profiles (non-ideal) --- 240
APPENDIX K - Density profiles (non-ideal) --- 244
APPENDIX L - Extra Validation temperatures data --- 248
vii
LIST OF TABLES
Table 1-1 - Heavy Oil Deposits in Utah (Speight 2013b) ... 4
Table 2-1 - Experimental results (Das and Butler 1994) ... 25
Table 2-2 - Total Oil production - Free fall and Diluted Oil (Haghighat and Maini 2013) ... 45
Table 2-3 - Diffusivity of C3H8 in heavy oil at different diffusion time 23.9oC (Tharanivasan et al. 2006) .. 57
Table 2-4 - Diffusion coefficient of solvents in oils (Wen et al. 2013) ... 67
Table 2-5 - Results for multi-component fluid at 422k & 15 MPa (Leahy-Dios and Firoozabadi 2007) ... 83
Table 4-1- Sample calculation for non-ideal density ... 104
Table 4-2 - Sample calculations for ideal density ... 107
Table 4-3 - Notation for bitumen height analysis ... 109
Table 4-4 - Difference in left tube pixel points from time t = 0 minutes and t = 660 minutes ... 114
Table 4-5 - Non-ideal mixing specific volume expressions ... 120
Table 4-6 - Full list of solubility limits at different conditions ... 123
Table 5-1 - Tabulated values for butane solvent decrease data (t = 0 and t ≠ 0) ... 129
Table 5-2 - Tabulated values for bitumen increase data (t = 0 and t ≠ 0) ... 133
Table 5-3 - SRT graph slope comparisons between James (2009) and this work ... 135
Table 5-4 - Ideal and non-ideal mixing density results ... 139
Table 5-5 - Non-ideal mixing specific volume expressions ... 140
Table 5-6 - Diffusivity functions for all temperatures (ideal and non-ideal) ... 141
Table 5-7 - Table showing % difference between ideal and non-ideal diffusion values ... 147
Table 5-8 - Validation data for butane solvent height decrease at all temperatures ... 164
Table 5-9 - Butane solvent height decrease data (28.50oC) ... 166
Table 5-10 - Comparison of 28.50oC butane decrease graph data to other temperatures ... 169
Table 5-11 - Bitumen height increase graph data (28.50oC) ... 170
Table 5-12 - Comparison of 28.50oC bitumen increase graph data to other temperatures... 172
Table 5-13 - Non-ideal mixing expression (28.50oC) vs other temperatures ... 174
Table 5-14 - Diffusivity results (28.50oC)... 174
Table 5-15 - Comparison of ideal/non-ideal diffusivity function deviation for 28.50oC ... 177
Table 5-16 - Standard deviations for temperature logging of all experimental runs ... 182
viii
LIST OF FIGURES
Figure 1-1 - Resource Pyramid focusing on Unconventional resources (Rajnauth 2012) ... 2
Figure 1-2 - Global heavy oil resources (Law 2011) ... 3
Figure 1-3 - Orinoco Belt Venezuela (EIA 2014a) ... 4
Figure 1-4 - Schematic of Alberta’s Oilsand deposits (Lunn 2013) ... 6
Figure 1-5 - Schematic of Alberta’s and Saskatchewan’s deposits (Nasr and Ayodele 2005) ... 7
Figure 1-6 - Typical CSS system (Lunn 2013) ... 10
Figure 1-7 - Typical SAGD set-up (Lunn 2013) ... 12
Figure 1-8 - Cross sectional schematic of SAGD process (Irani and Gates 2013) ... 13
Figure 1-9 - Cross section of VAPEX system (James 2009) ... 15
Figure 1-10 - N-Solv Process (Stickler 2009) ... 17
Figure 2-1 - Schematic of experimental Set-up (Jiang 1997) ... 26
Figure 2-2 - Change in VAPEX position with time for one of the models (James and Chatzis 2004) ... 30
Figure 2-3 - VAPEX interface and concentration profiles (James et al. 2007) ... 34
Figure 2-4 - VAPEX interface advancement for 92cm High Model (James and Chatzis 2005) ... 35
Figure 2-5 - Dispersion coefficient vs. permeability for all runs (El-Haj 2007) ... 37
Figure 2-6 - Mass transfer coefficient for various models (James and Chatzis 2007) ... 38
Figure 2-7 - Graph of Dispersion coefficient vs mass fraction (Abukhalifeh et al. 2009) ... 40
Figure 2-8 - Diffusion coefficient vs. Kinematic viscosity (Okazawa 2009) ... 42
Figure 2-9 - Oil production via gravity drainage w/wo dissolution (Haghighat and Maini 2013) ... 44
Figure 2-10 - Calculated viscosities vs. observed viscosities (Shu 1984) ... 47
Figure 2-11 - Comparison of viscosity predictions with various correlations (Yazdani and Maini 2010) ... 50
Figure 2-12 - Concentration dependence of viscosity and diffusivity (Das and Butler 1996) ... 51
Figure 2-13 - Concentration dependency of viscosity & diffusivity butane (Yazdani and Maini 2009) ... 52
Figure 2-14 - Pressure decay schematic (Zhang et al. 2000) ... 55
Figure 2-15 - Pressure decay graph (Zhang et al. 2000) ... 55
Figure 2-16 - Schematic diagram of the pressure cell set-up (Behzadfar and Hatzikiriakos 2014) ... 58
Figure 2-17 - DPDSA set-up (Yang and Gu 2006) ... 59
Figure 2-18 - CT images of a sand pack sample (Luo and Kantzas 2008) ... 63
Figure 2-19 - CT Scan at different times (Luo and Kantzas 2008) ... 63
Figure 2-20 - CT converted concentration profile at time 667 minutes (Luo and Kantzas 2008) ... 64
Figure 2-21 - Diffusivity in Heavy oil for bulk fluids and porous fluids (Luo and Kantzas 2008) ... 64
Figure 2-22 - Typical NMR spectra for bitumen, pure solvent and a mixture (Wen et al. 2013) ... 66
Figure 2-23 - Spectra change during diffusion (Wen et al. 2013) ... 67
Figure 2-24 - Solvent decrease (experimental vs model results) (James 2009) ... 78
Figure 2-25 - BIP for bitumen vs. temperature (Saryazdi et al. 2013) ... 81
Figure 2-26 - Infinite dilution diffusivity: experimental & computed (Leahy-Dios and Firoozabadi 2007) 84 Figure 2-27 - Ideal/non ideal mixture density vs. solvent volume fraction at 40oC (Etminan et al. 2011) . 85 Figure 3-1 - Simplified schematic of experiment ... 88
ix
Figure 3-2- Workflow diagram for experiment ... 89
Figure 3-3 - Butane loading set-up ... 90
Figure 3-4 - Schematic of butane side ... 91
Figure 3-5 - Butane side of one-dimensional diffusion experiment ... 92
Figure 3-6 - Schematic of bitumen side ... 93
Figure 3-7 - Schematic of bitumen loading ... 94
Figure 3-8 - Magnified schematic of bitumen loading ... 95
Figure 3-9 - Bitumen side ... 96
Figure 3-10 - Bitumen height pictures during vacuum creation (A) and at start of experiment (B) ... 99
Figure 3-11 - A side by side look at both L-tubes (during A and B) and R-tubes (during A and B) ... 99
Figure 3-12 - Full experimental set-up ... 100
Figure 3-13 - Detailed experimental schematic (butane and bitumen side together) ... 100
Figure 4-1 - Picture of bitumen height at t = 0 minutes ... 109
Figure 4-2 - Picture of bitumen height at time t = 660 minutes (11 hours) ... 113
Figure 4-3 - Bitumen mixture specific volume plots for ideal and non-ideal mixing ... 117
Figure 4-4 - Non-ideal mixing coefficients at varying temperatures ... 120
Figure 4-5 - Solvent solubility at varying experimental conditions ... 122
Figure 5-1- Butane height decrease vs. time (all temperatures) ... 127
Figure 5-2 - Butane solvent decrease vs SRT for t = 0 and t ≠ 0 ... 128
Figure 5-3 - Butane height decrease vs. SRT (all temperatures) ... 130
Figure 5-4- Bitumen height increase vs. time (all temperatures) ... 131
Figure 5-5 - Bitumen increase vs SRT for t = 0 and t ≠ 0 ... 132
Figure 5-6 - Bitumen height increase vs. SRT (all temperatures) ... 134
Figure 5-7 - Mixture densities vs. time (all temperatures) ... 136
Figure 5-8 - Mixture density vs. solvent mass fraction (40.00oC) ... 137
Figure 5-9 - Mixture density vs. solvent mass fraction (36.75oC) ... 137
Figure 5-10 - Mixture density vs. solvent mass fraction (33.50oC) ... 138
Figure 5-11 - Mixture density vs. solvent mass fraction (30.25oC) ... 138
Figure 5-12- Mixture density vs. solvent mass fraction (27.00oC) ... 139
Figure 5-13 - Butane solvent mass fraction profile at 27.00oC (non-ideal) ... 142
Figure 5-14 - Bitumen density profile at 27.00oC (non-ideal) ... 144
Figure 5-15- Solvent diffusivity functions at 27.00oC ... 145
Figure 5-16 - Solvent diffusivity functions at 33.50oC ... 146
Figure 5-17 - Solvent diffusivity functions at 40.00oC ... 146
Figure 5-18 - Ideal and non-ideal butane solvent mass fraction profile at 27.00oC ... 148
Figure 5-19 - Ideal and non-ideal butane solvent mass fraction profile at 33.50oC ... 149
Figure 5-20 - Ideal and non-ideal butane solvent mass fraction profile at 40.00oC ... 150
Figure 5-21 - Ideal and non-ideal bitumen density profile at 27.00oC ... 151
Figure 5-22 - Ideal and non-ideal bitumen density profile at 33.50oC ... 152
Figure 5-23 - Ideal and non-ideal bitumen density profile at 40.00oC ... 152
x
Figure 5-24- Non-ideal diffusivity functions at all temperatures ... 153
Figure 5-25 - Solvent mass fraction profile at 60 minutes for all temperatures ... 155
Figure 5-26 - Bitumen density profile at 60 minutes for all temperatures ... 156
Figure 5-27 - Solvent mass fraction profile at 2000 minutes for all temperatures ... 157
Figure 5-28 - Bitumen density profile at 2000 minutes for all temperatures ... 157
Figure 5-29 - Solvent mass fraction profile at 4500 minutes for all temperatures ... 158
Figure 5-30 - Bitumen density profile at 4500 minutes for all temperatures ... 159
Figure 5-31 - Predicted and experimental change in solvent height vs time at 33.50oC ... 160
Figure 5-32 - Predicted and experimental change in solvent height vs SRT at 33.50oC ... 161
Figure 5-33- Predicted and experimental change in solvent height vs time at 27.00oC ... 162
Figure 5-34 - Predicted and experimental change in solvent height vs SRT at 27.00oC ... 162
Figure 5-35 - Predicted and experimental change in solvent height vs time at 40.00oC ... 163
Figure 5-36 - Predicted and experimental change in solvent height vs SRT at 40.00oC ... 163
Figure 5-37 - Butane solvent decrease vs SRT for t = 0 and t ≠ 0 at 28.50oC ... 166
Figure 5-38 - Butane solvent height decrease vs. SRT (28.50oC) ... 167
Figure 5-39 - Butane solvent decrease DE predicted vs experimental values (28.50oC) ... 168
Figure 5-40 - Bitumen height increase vs SRT for t = 0 and t ≠ 0 at 28.50oC ... 169
Figure 5-41 - Bitumen height increase vs SRT at 28.50oC ... 170
Figure 5-42 - Bitumen increase DE predicted vs experimental values (28.50oC) ... 171
Figure 5-43 - Bitumen mixture specific volume plots for ideal and non-ideal mixing 28.50oC ... 173
Figure 5-44 - Experimental and DE predicted diffusivity functions 28.50oC (ideal/non-ideal) ... 175
Figure 5-45- Comparison of 28.50oC diffusivity function to other temperature (ideal) ... 176
Figure 5-46 - Comparison of 28.50oC diffusivity function to other temperature (non-ideal) ... 176
Figure 5-47 - Comparison of 28.50oC solvent profile to other temperatures (non-ideal) ... 178
Figure 5-48 - Comparison of 28.50oC density profile to other temperatures (non-ideal) ... 179
Figure 5-49 - Predicted and experimental change in solvent height vs SRT at 28.50oC ... 180
Figure 5-50 - Predicted and DE predicted change in solvent height vs SRT at 28.50oC ... 180
Figure 5-51 - Graph of temperature vs time for experiment (27.00oC bitumen and 24.00oC butane) .... 183
Figure 5-52 - Graph of temperature vs time for experiment (33.50oC, 36.75oC and 40.00oC bitumen) .. 184
Figure 5-53 - Smooth pressure decline at 33.50oC ... 185
Figure 5-54 - Pressure spikes at 27.00oC ... 186
Figure 5-55 - Pressure spikes at 40.00oC ... 187
xi
APPENDIX
Appendix - A1: Viscosity correlations ... 199
Appendix - A2: Bitumen Assay ... 200
Appendix - B3: Two sided butane set-up ... 203
Appendix - B4: PVTSim butane solvent temperature vs vapour pressure graph ... 205
Appendix - C5: Starting experimental parameters ... 207
Appendix - D6: Sample calculation for non-ideal density ... 208
Appendix - D7: Sample calculations for ideal density ... 212
Appendix - E8: Bitumen height analysis ... 215
Appendix - F9: Bitumen mixture specific volume plots for ideal and non-ideal mixing 40.00oC... 216
Appendix - F10: Bitumen mixture specific volume plots for ideal and non-ideal mixing 36.75oC... 216
Appendix - F11: Bitumen mixture specific volume plots for ideal and non-ideal mixing 33.50oC... 217
Appendix - F12: Bitumen mixture specific volume plots for ideal and non-ideal mixing 30.25oC... 217
Appendix - F13: Bitumen mixture specific volume plots for ideal and non-ideal mixing 27.00oC... 218
Appendix - G14: Butane solvent height decrease graph data (all temperatures) ... 220
Appendix - G15: Bitumen height increase graph data (all temperatures) ... 221
Appendix - J16: Butane solvent mass fraction profile at 30.25oC (non-ideal) ... 240
Appendix - J17: Butane solvent mass fraction profile at 33.50oC (non-ideal) ... 241
Appendix - J18: Butane solvent mass fraction profile at 36.75oC (non-ideal) ... 242
Appendix - J19: Butane solvent mass fraction profile at 40.00oC (non-ideal) ... 243
Appendix - K20: Bitumen density profile at 30.25oC (non-ideal) ... 244
Appendix - K21: Bitumen density profile at 33.50oC (non-ideal) ... 245
Appendix - K22: Bitumen density profile at 36.75oC (non-ideal) ... 246
Appendix - K23: Bitumen density profile at 40.00oC (non-ideal) ... 247
Appendix - L24: Predicted and experimental change in solvent height vs SRT at 30.25oC ... 248
Appendix - L25: Predicted and experimental change in solvent height vs SRT at 36.75oC ... 249
Appendix - L26: Validation data for bitumen height increase at all temperatures ... 250
xii
NOMENCLATURE
Symbol Description Units
A Area cm2, m2
C Concentration g/mol
D Diffusivity cm2/s, m2/s
d Depth cm, m, mm, μm
g Gravitational constant m/s2, cm/s2
h Height cm, m, mm
k Permeability Darcy, μm2
L Length cm, m, mm
m Mass g, kg
ṁ Mass flow rate g/min
N Flux g/cm2.s, kg/m2.s
P Pressure kPa, MPa, psi
PV Pore volume cm3
Q Volumetric flow rate cm3/min, barrels/day (bbp)
S Saturation %
t Time day, hour, min, s
U Velocity m/s, cm/s
V Volume m3 , cm3, barrels, bb
Vm Mass average velocity m/s, cm/s
v Volume fraction
w Width M, cm
Greek
δ Depth of the draining live oil μm, mm, pores
∆ change
φ Porosity %
μ Viscosity mPa.s, cP
ρ Density kg/m3, g/cm3
σ Surface tension N/m, dyne/cm
ω Mass fraction
xiii
Subscripts
b Bitumen
eq Equivalent
g, G Gas
i Interface or nodal position
l, L Liquid
lo Live oil
m Cementation factor
max Maximum
min Minimum
mix Mixture
n Exponent
o Oil
p Pore
s Solvent
v Vapor
wb Water bath
sys System
Superscript
* Solubility limit, i.e. ωs*
i Interfacial
dr Drainage
x X-direction
Acronyms/Definitions
OOIP Original oil in place
EIA Energy Information Administration
DE Design Expert
DE Design Expert
Diffusivity Measure of the magnitude of diffusion (usually varies) Diffusivity function Mathematical expression for diffusion
Diffusion coefficient Measure of the magnitude of diffusion (usually constant) Diffusion value Numerical values obtained from diffusivity function One-dimensional diffusivity Diffusion occurring in one direction and no outside impact
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 1
1) INTRODUCTION
Canada has the third largest World oil reserves behind Saudi Arabia and Venezuela with 174 billion barrels (bbls) in proven reserves (CAPP 2014), 97% are oil sands. These oil sands are typically extracted through mining or in-situ methods. The in-situ methods predominantly comprise of some form of thermal stimulation. One of such technology is Steam Assisted Gravity Drainage (SAGD).
However, thermal methods are not always ideal for in-situ formations, and can be replaced with more energy efficient mass transfer methods (Vapor Extraction VAPEX or N-Solv). Although some research has been done on the implementation of this technology, the key diffusion mechanism at the crux of the technology is yet to be fully understood.
Oil sands are a natural mixture of sand water, clay and bitumen. Bitumen is oil that is too heavy or thick to flow, or be pumped out thereby needing a form of heat stimulation or dilution (VAPEX or N-Solv) for it to flow. It typically possesses an API of less than 10o, viscosity of greater than 10000 cp and density of about 1000 kg/m3 (Law 2011). SAGD technology is energy intensive, produces adverse environmental effects, and is not economical for a number of heavy oil rich areas.
The purpose of this research is to evaluate the one – dimensional diffusivity of butane solvent in bitumen at varying temperature conditions, while accounting for changes in concentration and non-ideal mixing of hydrocarbons. Diffusivity function is a mathematical expression for diffusion as a function of another term (mass fraction in this work). It should be noted that one- dimensional diffusivity represents diffusion occurring in one direction and no outside impacts.
The difference between diffusivity functions and diffusion values should also be noted. Diffusion values are the numerical values obtained when mass fraction (ωs) is substituted into the diffusivity function and computed. The study also intends to find a correlation that will predict the diffusivity within the given temperature range.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 2
1.1. World Heavy Oil Overview
Heavy oil and bitumen are types of crude oil that fall under the unconventional resource umbrella. The key difference between conventional and unconventional resources is that conventional resources are able to flow in their natural state through production conduits (wells) while being economical for production (Vassilellis 2009). Unconventional resources are unable to be produced at economical rates without assistance from massive stimulation treatments or special recovery processes (Haskett and Brown 2005). Figure 1-1 shows a resource pyramid for conventional and unconventional resources. It should be noted that in this current day Gas Hydrates technology is still in the works while Shale Oil technology is fully commercial.
Figure 1-1 - Resource Pyramid focusing on Unconventional resources (Rajnauth 2012)
Most heavy oil reservoirs originated as conventional oil that formed in deep formations that later migrated to the surface. They were then degraded by bacteria and weathering leading to the escape of the lightest hydrocarbons. As a result, heavy oil is deficient in hydrogen and high in carbon, sulphur and heavy metals. This also leads to heavy oil having API gravity of below 22.3o and viscosity of over 1,000 cP (Speight 2013a).
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 3 There are conflicting definitions as to what is heavy oil versus bitumen. Technically bitumen is a crude grade that has an API of less than 10o and viscosity of greater than 10,000 cP (Law 2011).
It also exists in solid and semi-solid states in the reservoir. Heavy oil is a crude grade with an API of 10o – 22.3o and viscosity of 100 – 10,000 cP. It is predominantly found in a highly viscous liquid state and can also be referred to as ‘extra heavy oil’. For the purpose of this research, the definitions of heavy oil will encompass normal bitumen, heavy oil, and extra heavy oil.
As profitable as heavy oil can be, there are smaller profit margins achieved with its production when compared to conventional oil. This is mainly due to higher production costs, upgrading costs and lower market price for heavier crude oils.
From a location stand point heavy oil is predominantly found in Canada and Venezuela. Other countries include the US, Russia, Brazil, and China. Figure 1-2 shows a diagram of OOIP (Original Oil in Place) heavy oil resources around the world.
Figure 1-2 - Global heavy oil resources (Law 2011)
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 4 Heavy oil deposits in Venezuela are the largest in the world with 298 billion barrels of proved reserves (EIA 2014b). The vast majority are located in the Orinoco heavy oil belt. Figure 1-3 shows a diagram of the heavy oil deposits. The U.S. Energy Information Administration (EIA) estimates Venezuela produced 2.49 million barrels per day (bpd) of petroleum and other liquids in 2013.
Figure 1-3 - Orinoco Belt Venezuela (EIA 2014a)
Major heavy oil deposit in the US, are found mostly in the Uinta Basin, Utah. These include the Sunnyside, Oil sand triangle, Peor Springs, Asphalt Ridge and Sundry deposits. Table 1-1 shows a list of these deposits along with their reserves.
Table 1-1 - Heavy Oil Deposits in Utah (Speight 2013b)
Deposit Known Reserves (bbl×106)
Additional Probable Reserves (bbl×106)
Sunnyside 4,400 1,700
Tar Sand Triangle 2,500 420
PR Spring 2,140 2,230
Asphalt Ridge 820 310
Circle Cliffs 590 1,140
Other 1,410 1,530
Total 118,060 7,330
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 5 Brazil possessed 13 billion bbls of proven oil reserves as of January 2013 (EIA 2013). The offshore Campos and Santos Basins, located off the country's southeast coast, hold the vast majority of Brazil's proved reserves, 90% of which are heavy oil. In China four natural heavy oil accumulations are located in the Junggar basin with resources of about 1.6 billion bbls (USGS 2010). In Russia, large resources are present in the east Siberia platform inside the Tunguska basin. The area is conservatively estimated to contain 51 billion bbls of oil reserves (USGS 2010). The area, however, is very remote and unlikely to be exploited in the near future due to Russia’s large conventional oil and gas resources.
1.2. Canadian Perspective
Canada, with 1.7 trillion bbls of original oil in place OOIP, is estimated to have 168 billion barrels of oil sands (Alberta Energy 2010). The main locations are in the provinces of Alberta and Saskatchewan.
Alberta’s heavy oil deposits are located in Athabasca, Peace River and Cold Lake. These deposits combined, occupy an area of 54,000 square miles (139,860 km2) (Alberta, Energy Resources Conservation Board 2013). This is just over half the size of the entire United Kingdom (243,610 km2). Figure 1-4 shows a schematic of Alberta’s heavy oil deposits.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 6 Figure 1-4 - Schematic of Alberta’s Oilsand deposits (Lunn 2013)
Athabasca deposits are the largest and most accessible reserve of bitumen. It underlies a 16000 m2 area and contains 812 billion bbls of bitumen in place (Alberta, Energy Resources Conservation Board 2013). The rich bitumen accumulation is covered by overburden between 5 – 100 m typically with an average pay zone of 20 m. 7% of the deposits lie under less than 5 – 100 m of overburden making it accessible to surface mining techniques and 33 billion barrels are estimated to be recoverable by mining methods (Speight 2013b). Also worth mentioning is the Wabasca oil sand deposit, which is usually indicated as part of the Athabasca reserve. It is estimated to have 42.5 billion bbls of OOIP and at 490 – 1500 ft (149.35 – 457.2 m) of over- burden (Speight 2013b) it is only recoverable by in-situ methods.
Cold lake deposits are the second largest of the three and have four separate reservoir deposits – one each in McMurray, Clearwater, Lower Grand Rapids and Upper Grand Rapids. Given depth varies from 984 – 1969 ft (299.92 – 600.15 m) with surface mining not possible, the deposits are
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 7 suitable for in-situ methods. Cold lake holds approximately 178 billion barrels of oil reserves (Speight 2013b).
The Peace River deposit has bitumen at a depth of 1000 to 2500 ft (304.8 – 762 m) in the Blue- sky and Gething formation. With an area of 3000 m2 there are 71.7 billion barrels of bitumen in place.
In Saskatchewan, oil sands are found in McMurray formation sediments equivalent to those of the Athabasca deposits. Bitumen-bearing sands in the McMurray formation extend from Alberta into Saskatchewan and are estimated to contain 20.5 billion bbls OOIP (Speight 2013b). These oil sands are known as ‘shallow in-situ oil sands’. This is because there is substantial difficulty in driving and producing the bitumen while managing water flows in the reservoir. They are found between 75 – 200 m making extraction with both mining and normal in-situ methods difficult.
VAPEX technology currently represents the best opportunity to recover these reserves. Figure 1- 5 shows a schematic of Alberta’s and Saskatchewan’s deposits.
Figure 1-5 - Schematic of Alberta’s and Saskatchewan’s deposits (Nasr and Ayodele 2005)
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 8
1.3. In-situ Extraction Methods for Bitumen
Extraction methods can be classified under mining or in-situ methods. Mining methods are typically used when the heavy oil is close enough to the surface with about 75 m of over-burden (Schramm et al. 2010) . This process typically gives a recovery factor of 95% or greater. In-situ methods are used when depths are greater than 200 m of over-burden with recovery factors ranging from 50% – 60% (Speight 2013a) 75 – 200 m fall under ‘shallow in-situ’ reserves as stated above.
Surface Mining Methods - This is the excavation (surface or sub-surface) of petroleum resources for subsequent removal of the bitumen via washing, floatation or restoring treatments. Athabasca deposits are the only ones that are shallow enough to be extracted via mining methods. Of Alberta’s 167 billion bbls of oil reserves 81% are recoverable via in-situ methods while the remaining 19% are close enough to the surface and can be mined (Alberta, Energy Resources Conservation Board 2013). About 1.2 million bbls per day are produced via surface mining methods, almost exclusive to Canada (Vassilellis 2009).
A typical mining operation removes one and a half ton of over-burden, mines two tonnes of oil sands and processes it to yield one bbl of bitumen after extraction. With increasing depth, the grade of oil sand decreases and additional tonnes must be mined and processed to yield the same amount of bitumen. Commercial mining operations therefore, have economic depth limits and economic grade limits. Both of these are dictated by the trade-off between mining and processing costs vs. the value of the bitumen (Schramm et al. 2010).
After extraction the bitumen is separated from the sand through the ‘Hot Water Process’. This utilizes linear and the non-linear variations of bitumen density and water density, respectively, with temperature, so that the bitumen that is heavier than water at room temperature becomes lighter than water at 80°C (Speight 2013b). To date, hot water processes are still the only commercially viable way to separate bitumen from the sand.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 9 In-situ methods – In Latin terms, the word in-situ stands for ‘in position’. When this is used in heavy oil production scenarios, it stands for extraction done underground. This means extraction is occurring in the place where the oil resources are located. Deep oil sand reservoirs (greater than 200 m in depth) lend themselves to this form of extraction.
In-situ extraction methods encompass thermal (Cyclic processes, SAGD and Hybrid processes) and non-thermal methods (VAPEX and N-Solv). For this thesis, Solvent technology will encompass VAPEX and N-Solv methods. Thermal methods, as the name suggest, requires a form of heat stimulation for extraction. It is, therefore, water and energy intensive, while still being the most economical of all the processes. Non–thermal methods usually use solvents to dilute the bitumen, reducing its viscosity and leading to extraction. This thesis will exclusively cover in-situ methods that are used for bitumen where the viscosity reduction is first required.
1.3.1. Cyclic Steam Stimulation
Cyclic Steam Stimulation (CSS) involves alternating between injecting the well with steam and producing the same well using the condensed steam (Speight 2013a). This cycling occurs with a single vertical well which serves as both an injector and producer well (this is sometimes referred to as “huff and puff”). An alternative incorporates steam drive between injectors and producers. These processes originally depended on vertical wells, but a combination of vertical and horizontal wells are now used. CSS can be used for bitumen, heavy and extra heavy oil grades. Figure 1-6 shows a schematic of a typical CSS system.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 10 Figure 1-6 - Typical CSS system (Lunn 2013)
As shown above, CSS is a three–stage process: first, high–pressure steam is injected into the vertical well for a period of time; second, the reservoir is shut in to soak; and third, the well is put into production. Production rate decline prompts the start of another cycle of steam injection.
The injection–production cycle is repeated a number of times over the life of the well. The time required to steam and produce the wells varies from well to well with each cycle, typically, between 6 and 18 months.
Steam generated is injected into the formation through the wellbore at a temperature of about 300°C and pressures averaging 11,000 kPa. This pressure is sufficient to cause parting of the unconsolidated oil sands formation, creating paths for fluid flow (Speight 2013a). The recovery factors for CSS range from about 20 – 25% of original bitumen–in–place.
Typical steam-to-oil ratios, the major economic factor, are 3:1 – 4:1. Although CSS is characterized by higher steam–oil ratios than with SAGD, the quality of steam used is much lower and requires less energy to produce. In CSS operations, natural gas requirements can be
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 11 met through produced solution gas from the process. The same is not applicable SAGD operations, because of the low amount of solution gas produced.
Bitumen produced by CSS tends to have a higher API gravity and is less viscous; therefore, diluent costs are reduced when compared with bitumen produced by the SAGD process. A key focus in a CSS operation is to increase the total recovered bitumen by increasing the quantity of bitumen recovered in each cycle and increasing the number of cycles for which bitumen recovery is economical. The steam–oil ratio, and therefore, gas costs for steam generation, is typically at its lowest point during early cycles. After this it begins to rise until the point at which bitumen production is no longer economical and the well is abandoned (Speight 2013b).
The CSS process is a well-developed process; the major limitation is its unfavorable recovery rate (usually less than 20%) of the initial oil-in-place. The process is particularly effective in reservoirs with limited vertical permeability and is best suited to operations in the Cold Lake area and the Peace River heavy oils.
1.3.2. SAGD
Steam-assisted gravity drainage (SAGD) is currently the most commercially successful heavy oil in-situ extraction method in Canada. Alberta produces approximately 1.4 million bbls per day of heavy oil via this technology (Alberta, Energy Resources Conservation Board 2013). The key element of SAGD is that the two wells need to be in parallel and horizontal form. The development of horizontal drilling in 1992 was a key break-through that led to the commercialization of SAGD technology. Figure 1-7 shows a schematic of a typical SAGD set- up.
This technology involves drilling two parallel horizontal wells along the reservoir itself. The top well is known as the injector well and used to inject hot steam into oil sands. The bottom well is known as the producer well and used to produce the oil and pump it up to the surface. Steam from the injector well builds up a steam chamber in the reservoir. Once this steam chamber reaches the bitumen source, it heats up the immobile bitumen leading to viscosity reduction of
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 12 the oil. With viscosity reduced by 10,000 cP, the oil becomes mobile, and drains into the producer well before being pumped to the surface.
Figure 1-7 - Typical SAGD set-up (Lunn 2013)
Figure 1-8 represents a cross sectional schematic of the SAGD process. Section A represents the initial circulation phase, where thermal communication is established between the wells. Section B is the early production stage in which the steam chamber has yet to come in contact with the oil formation (cap-rock). Section C is the lateral growth stage, where steam has fully come in contact with the oil and reduced its viscosity, leading to production. Some by-products of the production include non-condensable gases like methane.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 13 Figure 1-8 - Cross sectional schematic of SAGD process (Irani and Gates 2013)
Even though the injection and production wells can be very close (between 5 – 7 m), the mechanism causes the steam-saturated zone to rise to the top of the reservoir, expand gradually sideways, and eventually allow drainage from an increase area. Operating the production and injection wells at approximately the same pressure as the reservoir pressure eliminates viscous fingering and coning processes, and also suppresses water influx or oil loss through permeable streaks (Speight 2013a, Speight 2013b) . This keeps the steam chamber interface relatively sharp and reduces heat losses considerably. Injection pressures are much lower than the fracture gradient, which means that the chances of breaking into a thief zone, an instability problem which plagues all high-pressure steam injection processes, are essentially zero.
Heat losses and deceleration of lateral growth mean that there is an economic limit to the lateral growth of the steam chamber. This limit is thought to be a chamber width of four times the vertical zone thickness. For thinner zones, the horizontal well pairs would need to be put in closer proximity leading to cost increase as well as lower total resource per well pair.
Consequently, the zone thickness limit (net pay thickness) must be defined for all reservoirs.
SAGD is widely known to recover 50 – 70% of OOIP (Mukhametshina et al. 2014) and is therefore the most efficient thermal recovery method. The key benefits of the SAGD process are an improved steam-oil ratio and high ultimate recovery (on the order of 60 – 70%). Key issues still troubling the technology relate to low initial oil rate, artificial lifting of bitumen to the
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 14 surface, horizontal well operation, and the extrapolation of the process to reservoirs having low permeability, low pressure, or bottom water (aquifers). The cost of heat is also still a major economic constraint on all thermal processes. Currently, steam is generated with natural gas, and when the cost of natural gas rises, operating costs rise considerably.
1.3.3. VAPEX
VAPEX (Solvent Vapor Assisted Petroleum Extraction) technology can be summarized as SAGD technology with solvent vapor replacing the steam used. The solvent vapor ranges from ethane, propane to butane depending on circumstances. The key is that this solvent, given its significantly lower viscosity, would dilute the highly viscous bitumen leading to overall viscosity reduction and therefore the flow of oil. The bitumen viscosity reduction factor using these solvents is not as much as that seen in thermal-based methods, but the hope is that it becomes comparable enough for the technology to have a viable future.
The key parameter to this dilution occurring is mass transfer of the solvent into the bitumen through diffusion. The more soluble the solvent is in the bitumen, the more diffusion occurs and subsequent viscosity reduction of the bitumen.
Parallel horizontal wells are drilled with about 15 ft (4.57 m) of vertical separation in similar fashion to SAGD technology (Speight 2013b). The injected vapor forms a vapor chamber (analogous to SAGD’s steam chamber), through which the solvent travels to the immobile oil face where it diffuses into the immobile liquid. The viscosity reduced oil become mobile and drains to the production well via gravity drainage, where it is pumped to the surface. Figure 1-9 shows a cross sectional area schematic of a typical VAPEX system.
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 15 Figure 1-9 - Cross section of VAPEX system (James 2009)
VAPEX can be applied in paired horizontal wells, single horizontal wells, or a combination of vertical and horizontal wells. The mechanism of the VAPEX process is essentially the same as for the SAGD as said earlier, as is the configuration of the wells. The key benefits are claimed to be significantly lower energy costs, the potential for in-situ upgrading and the application to thin reservoirs with bottom water or reactive mineralogy. VAPEX also does not require water processing or recycling, offers lower carbon dioxide emissions and can be operated at deposit temperatures with no loss of heat. From a numbers perspective, VAPEX capital costs are about 75% of SAGD costs and 50% of SAGDs operating costs (Vargas-Vasquez and Romero-Zeran 2007). Research carried out thus far suggests that up to 90% of the solvent used can be recovered and recycled, offering the potential for dramatic cost savings over other extraction methods (Speight 2013b). Also, more wells are needed to achieve similar production rates and rates of recovery compared to SAGD.
There is also the chance of asphaltene precipitation. This can be an advantage as it reduces the viscosity of the drained oil, but conversely some feel that the precipitated asphaltene may clog portions of the pore space leading to lower production rates (Speight 2013b). However James (2009) work proves otherwise because: the upgraded oil sells at a higher price and
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 16 environmental/energy implications required by surface upgrading of the oil are reduced if the asphaltene are left in-situ.
Because of the slow diffusion of gases and liquids into viscous oils, this approach, would insufficient for use in its own, unless use for less viscous oils. Preliminary tests indicate, however, that there are micro–mechanisms that act, which indicate that the VAPEX dilution process is not diffusion–rate limited. This means the process may be suitable for the highly viscous heavy oil and tar sands.
1.3.4. N-Solv
This is a patented in-situ technology (Nenniger and Dunn 2008) that uses warm solvent to extract bitumen from oil sands. The concept of N–Solv is a fairly new type of technology and relative to other in-situ techniques could be seen as a mix between VAPEX and SAGD technology. N–Solv uses the same kind of solvents used in VAPEX, but adds a much bigger thermal aspect compared to VAPEX.
This process uses the proven horizontal well technology developed for SAGD, but differs as it substitutes water (steam) for a warm solvent (propane/butane). This is injected as vapor and condenses at the immobile bitumen interface, washing the valuable compounds out of the bitumen. A key benefit of the process is that it produces lighter, partially upgraded and more valuable oil products. It may also recover more resource from each well at lower capital and operating costs than existing in–situ processes. Environmental benefits like reduced CO2
emissions are also present.
While in VAPEX viscosity reduction mainly comes from diluting the bitumen with solvent through diffusion, N–Solv involves viscosity reduction through dilution (diffusion and convective mixing) with solvent and raising the temperature of the bitumen. It is a best of both worlds scenario and as a result it gives a much faster viscosity reduction factors compared to VAPEX. Also the solvent vapor is intentionally kept at lower pressure (or higher temperature) to stop it condensing to liquid in VAPEX systems. N–Solv conversely requires the condensation of
Concentration Dependent Non-Ideal Diffusion in Hydrocarbon systems Page 17 the solvent vapor for its latent heat to be given off. The sensible heat given to the bitumen to aids viscosity reduction, however, James, (2009) proves that the solvent mixing plays the biggest role in speeding up the N–Solv process.
Like in SAGD and VAPEX, the vapor flows from the injection well to the colder perimeter of the chamber, where it condenses. This delivers heat and fresh solvent directly to the bitumen extraction interface. The extraction conditions are mild compared to in–situ steam processes, so the valuable components in the bitumen are preferentially extracted. Figure 1-10 shows a schematic of your typical N–Solv process
Figure 1-10 - N-Solv Process (Stickler 2009)
1.3.5. Hybrid Processes
These are processes that involve the simultaneous use of several technologies for extraction and these processes are seen to hold significant promise for recovering worldwide reserves (Speight 2013a).